Professional Documents
Culture Documents
Notes Installment 1
Notes Installment 1
Natural Response
A significant portion of these notes are concerned with the study of finite-
dimensional, linear time-invariant (LTI) systems. We will define this term
with more care in section 1.3.2. Such systems can be described by finite-
order linear constant coefficient differential equations. Such models are
widely applicable to physical systems. In this chapter, we will be primarily
concerned with the natural response of such models, which is defined as the
response which occurs solely from initial conditions with no other inputs.
The natural response is also known as the unforced response or characteristic
response. The model differential equation for such a system is homogeneous,
in that there is no forcing term.
There is a beautiful property of LTI systems: the homogeneous or natural
response can be very simply found. It is composed of weighted sums of
functions est , where s is possibly complex (or most generally such functions
multiplied by polynomials in the time variable t). This is a statement about
the solution of differential equations. However, it is a remarkable empirical
result that such differential equations well-describe many physical systems.
Said another way, the types of natural responses discussed below can be
easily observed in an experimental context, and in observations of many
physical phenomena. The natural response ties things together.
A further surprising result is that real-world systems are frequently able
to be represented in terms of very simple models of first- or second-order.
When higher-order models are required, these systems have responses com
posed of sums of first- and second-order responses. So it’s very worthwhile
to understand the building-block first- and second-order responses in depth.
This chapter is organized as follows: We present first-order systems, and
their natural response, starting with a mechanical example. The charac
The response decays to zero with increasing time if τ > 0; if the natural
response of a system always decays to zero with increasing time for any
initial conditions, we say that the system is stable. If the response goes
off to infinity with increasing time for some initial conditions, the system
is unstable.
The response (1.5) has the initial value y(0) = c. The graph of this
response is shown in Figure 1.1 for an initial condition c = 1 with the four
values τ = 2, 1, 0.5, 0.1. As you can see, τ represents the characteristic time
for the response to decay toward zero; smaller values of τ correspond with
faster responses.
Because such a response is widely applicable in real engineering systems,
we will take a bit of time to understand it in more depth. Your efforts here to
internalize an understanding of this response and its characteristics will pay
dividends throughout your engineering studies and practice. Specifically, in
an interval of one time constant, the response shown decays to a value of
0.37 times the value at the start of the interval. This is so because e−1 =
0.3679 ≈ 0.37. Since this response has an initial value of 1, the response
decays to a value of 0.37 in one time constant, 0.372 in two time constants,
and a value of 0.37n in n time constants. You should verify this result to
graphical accuracy for all four of the time constant values; they pass through
the dashed line y = 0.37 in an interval equal to τ . And, in 3 seconds, the
response for τ = 1 sec passes through a value of 0.373 ≈ 0.05; we would
2
The terms characteristic frequency and eigenvalue are equivalent, and will be used
interchangeably herein.
8 CHAPTER 1. NATURAL RESPONSE
1
0.9
y(t) 0.8
0.7 decreasing
0.6
0.5
=2 0.37
0.4
0.3 =1
0.2 = 0.5 3
0.37=0.05
0.1 = 0.1
0
0 0.5 1 1.5 2 2.5 3
t(sec)
say that this response has settled to within 5% in 3 time constants. How
many time constants would it take to settle to within 1%? (You should
be sure you can answer this before going on.) Meanwhile, in 3 seconds,
the τ = 0.1 sec response has passed through 30 time constants, and has a
value of e−30 = 9.4 × 10−14 . This is pretty close to zero, but in theory the
response never quite gets to zero, no matter how long you wait; it just keeps
decaying by further factors of 0.37.
So we see that the eigenvalue captures the time-scale of the first-order
response. This idea extends to the higher-order systems considered later. In
these more general cases, the eigenvalues may have imaginary components;
in the first-order case considered above they are pure real. Because of the
primary importance of the eigenvalues of a system, it is common in prac
tice to graphically plot the eigenvalue locations on a plane with horizontal
axis Re{s}, and vertical axis Im{s}. This complex plane is referred to as
the s-plane, and the eigenvalue locations are called poles. For example, in
the first-order system considered above, the eigenvalue is λ1 = −1/τ ; this
system is thus also said to have a pole at s = λ1 = −1/τ . In the complex
plane, poles are plotted as x’s; for the first-order system, the pole diagram
1.1. FIRST-ORDER SYSTEMS 9
Im{s}
decreasing
Re{s}
X
1
1 =
Consider the mechanical system shown in the picture of Figure 1.3 as used
in Lab 1. This consists of a spring-steel beam rigidly fixed at one end, and
attached to an air cylinder damper on the other end. We will consider this
as a translational system, with the point of translation corresponding to the
nearly straight line motion of the end of the beam where it joins the air
piston damper. The air piston damper3 consists of a graphite piston sliding
in a precisely fit glass cylinder as shown in Figure 1.4. The knob at the near
end controls an adjustable orifice to set the resistance to flow in and out of
the damper, and thereby set the damping coefficient.
Figure 1.5 shows experimental data taken from this system via videotap
ing at 20 frames per second, as well as data from a model adjusted to match
this response. The measured data points are shown in blue, with asterisks
at the data points taken every 1/20th of a second. The red curve is a plot
of the first-order response (1.5) with the parameters adjusted to reasonably
3
Also known as an Airpot, which is a trademark of the Airpot Corporation, Norwalk,
CT.
1.1. FIRST-ORDER SYSTEMS 11
and thus the time constant is τ = 1/1.65 = 0.61 sec. The initial condition
is c = 1.5 cm.
The first-order model (1.5) fits this response very well. The experimental
data is a bit noisy as might be expected. The primary noise source is that
the video camera frame rate is not very constant. This could be improved
with better video hardware, but is not important for this experiment.
The simplest lumped mechanical model which fits this response is the
first-order mechanical spring-damper system shown in Figure 1.6. Here we
assume that the link can only move in the x-direction. The cantilever beam
acts as a spring which is linear for moderate deflections. The spring con
stant k for this beam can be calculated from first principles. With this
calculated spring constant we can compute the damping coefficient equiva
lent b for the air piston damper.
As shown in the figure, the system consists of a spring and damper
attached to a rigid massless link. The link represents the connection between
the spring and damper, but contributes no dynamics of its own. The position
of the link is denoted as x. The zero of position is indicated in the figure
by the vertical line connecting to the arrow which indicates the direction
12 CHAPTER 1. NATURAL RESPONSE
1
Position (cm)
0.5
0
0 0.5 1 1.5 2 2.5 3
Time (sec)
of increasing x. This choice of zero accounts for the rest position (zero
force length) of the spring. The spring is moved by a force proportional to
motion in the x-direction, Fk = kx. The damper is moved by a force which
is proportional to velocity in the x-direction, Fb = b dx/dt.
Newton’s second law states that F = ma = mẍ, where F is the sum of
the forces acting on a mass. This relationship also applies to the massless
link, but since the link is massless, the forces must instantaneously sum to
zero. For any mass element, or massless assembly from a system, Newton’s
second law can be captured in the form of a free-body diagram. For this
system the free-body diagram appears as shown in Figure 1.7.
Summing forces acting on the link and applying Newton’s second law
yields the system equation of motion
dx
−Fk − Fb = −kx − b = 0. (1.7)
dt
The minus signs appear here for the forces Fk and Fb since they act on the
link in the −x-direction. The zero term on the right is due to the fact that
the link is massless. The governing differential equation can be rewritten as
b dx
+ x = 0, (1.8)
k dt
If we define τ = b/k, this is in the form of (1.1). The natural response is
thus as calculated in section 1.1, with its associated figures.
14 CHAPTER 1. NATURAL RESPONSE
k Fk
b
Fb
Systemcuthere
Forcesactingonelements
Figure 1.7: Free body diagram for massless link of first-order system.
We can calculate that for the dimensions of this beam k = 170 N/m.
With this value in hand the model damping coefficient is given by
Figure 1.9: Picture of brass flywheel being placed on top of shaft of rotational
system.
differential equation.
dω
J + bω = 0 (1.10)
dt
which can be rewritten in standard form as
J dω
+ω =0 (1.11)
b dt
This equation has the solution
which gives
θ(t) = θ(0) + τ c[1 − e−t/τ ]. (1.17)
that is, with τ = 0.1 sec, θ(0) = 1 rad, and c = 85 rad/sec. This allows us
to give the estimated velocity as a function of time as
ic +
ir +
C vc vr
R
The currents ir and ic must be equal and opposite so that their sum is
equal to zero, since current cannot accumulate at their common node. That
is, we must have
vr dvc
0 = ir + ic = +C . (1.20)
R dt
Recognize further that since the two elements are connected in parallel,
their voltages must be equal: vr = vc . You substitute this into (1.20), and
multiply through by R to find
dvc
RC + vc = 0. (1.21)
dt
If we define τ = RC, this is in the form of (1.1). The natural response is
thus as calculated in section 1.1, with its associated figures. Specifically, if
the initial voltage on the capacitor is defined as vc (0) = v0 , then the voltage
as a function of time varies as
For example, suppose that we set C = 100 µF and R = 1 MΩ. Then the
time constant is τ = 100 sec; it will take the capacitor voltage 100 seconds
to decay to 37% of its initial value.
1.1. FIRST-ORDER SYSTEMS 21
# 4B 2B
B 4 4A
4HERMAL RESISTANCE
TO OUTSIDE WORLD
heat flow through those sides is zero. The block is connected to the outside
ambient temperature via the thermal resistance Rb , such that
Ta − Tb
qb = . (1.24)
Rb
This resistance represents the flow of heat into the bulb as a linear function of
the temperature difference4 between the ambient and the bulb temperatures.
Setting equality between the last two equations gives
dTb Ta − Tb
Cb = . (1.25)
dt Rb
Now, it’s convenient to define a variable to represent the temperature dif
ference between the bulb and ambient: T ≡ Tb − Ta . Since the ambient
temperature is constant, dT /dt = dTb /dt. Making these substituations and
multiplying (1.25) through by Rb yields
dT
Rb Cb + T = 0. (1.26)
dt
If we define τ = Rb Cb , this is in the form of (1.1). The natural response is
thus as calculated in section 1.1, with its associated figures. Specifically, if
the initial temperature difference of the bulb is defined as T (0) = T0 , then
the temperature difference as a function of time varies as
If you want to convert back to the absolute temperature of the bulb, re
member that Tb = T + Ta .
Pa − Pt −ρgh
qt = = . (1.28)
R R
dh qt
= (1.29)
dt A
We multiply through by A and set equality between the last two equa
tions to give
RA dh
+ h = 0. (1.30)
ρg dt
If we define τ = RA/ρg, this is in the form of (1.1). The natural response
is thus as calculated in section 1.1, with its associated figures. Specifically,
1.2. SECOND-ORDER SYSTEMS 25
if the initial fluid height is defined as h(0) = h0 , then the fluid height as a
function of time varies as